Multi-objective Evolutionary Algorithm Research Articles

Parameter-less Pareto local search for multi-objective neural architecture search with the Interleaved Multi-start Scheme

With the emerging deployment of deep neural networks, such as in mobile devices and autonomous cars, there is a growing demand for neural architecture search (NAS) to automatically design powerful network architectures. It is more reasonable to formulate NAS as a multi-objective optimization problem. In addition to prediction performance, multi-objective NAS (MONAS) problems take into account other criteria like the number of parameters and inference latency. Multi-objective evolutionary algorithms (MOEAs) are the preferred approach for tackling MONAS due to their effectiveness in dealing with multi-objective optimization problems. Recently, local search-based NAS algorithms have demonstrated their efficiency over MOEAs for MONAS problems. However, their performance has been only verified on bi-objective NAS problems. In this article, we propose a local search algorithm for multi-objective NAS (LOMONAS), an efficient local search framework for solving not only bi-objective NAS problems but also NAS problems having more than two objectives. We additionally present a parameter-less version of LOMONAS, namely IMS-LOMONAS, by combining LOMONAS with the Interleaved Multi-start Scheme (IMS) to help NAS practitioners avoid manual control parameter settings. Experimental results from a series of benchmark problems in the CEC’23 Competition demonstrate the competitiveness of LOMONAS and IMS-LOMONAS compared to MOEAs in tackling MONAS within both small-scale and large-scale search spaces. Source code is available at: https://github.com/ELO-Lab/IMS-LOMONAS.

Multi-objective path planning for multi-UAV connectivity and area coverage

In this paper, we assume that a team of drones equipped with sensing and networking capabilities explore an unknown area via onboard sensors for surveillance, monitoring, target search or data collection purposes and deliver the sensed data to a ground control station (GCS) over multi-hop links. We propose a multi-drone path planner that jointly optimizes area coverage time and connectivity among the drones. We propose a novel connectivity metric that includes not only percentage connectivity of the drones to GCS, but also the maximum duration of consecutive time that the drones are disconnected from the GCS. To solve this optimization formulation, we propose a multi-objective evolutionary algorithm with novel operations. We use our solver to test single, two and many objective path planning problems and compare our Pareto-optimal solutions to benchmark weighted-sum based solutions. We show that as opposed to the single solution that weighted-sum methods provide based on prior information from the user, the proposed evolutionary multi-objective optimizers can provide a diverse set of solutions that cover a range of mission time and connectivity performance illustrating the trade-off between these conflicting objectives. The end-user can then choose the best path solution based on the mission priorities during operation.

Realtime optimization and management system (ROAM): A decision support system for digital agriculture systems

The growing disparity between food supply and demand requires innovative Digital Agriculture (DA) systems to increase farm sustainability and profitability. However, current systems suffer from problems of complexity stemming from the challenge of integrating diverse, often non-interoperable hardware and software components. In order to tackle these complexities to increase farm efficiency and understand the tradeoffs of these new DA innovations we developed Realtime Optimization and Management System (ROAM), which is a decision-support system developed to find a Pareto optimal architectural design to build DA systems. To find the Pareto optimal solution, we employed the Rhodium Multi-Objective Evolutionary Algorithm (MOEA), which systematically evaluates the trade-offs in DA system designs. Based on data from five live deployments at Cornell University, each DA design can be analyzed based on user defined objectives and evaluated under uncertain farming environments with ROAM. Paired with this, we develop a web interface that allows users to define personalized decision spaces and visualize decision tradeoffs. To help validate ROAM, it was deployed to a commercial farm where the user was recommended a DA architecture design method to increase farm efficiency. ROAM allows users to quickly make key decisions in designing their DA systems to increase farm profitability.

Comparison of MOEAs in an Optimization-Decision Methodology for a Joint Order Batching and Picking System

This paper investigates the performance of a two-stage multi-criteria decision-making procedure for order scheduling problems. These problems are represented by a novel nonlinear mixed integer program. Hybridizations of three Multi-Objective Evolutionary Algorithms (MOEAs) based on dominance relations are studied and compared to solve small, medium, and large instances of the joint order batching and picking problem in storage systems with multiple blocks of two and three dimensions. The performance of these methods is compared using a set of well-known metrics and running an extensive battery of simulations based on a methodology widely used in the literature. The main contributions of this paper are (1) the hybridization of MOEAs to deal efficiently with the combination of orders in one or several picking tours, scheduling them for each picker, and (2) a multi-criteria approach to scheduling multiple picking teams for each wave of orders. Based on the experimental results obtained, it can be stated that, in environments with a large number of different items and orders with high variability in volume, the proposed approach can significantly reduce operating costs while allowing the decision-maker to anticipate the positioning of orders in the dispatch area.

Constructing a MOEA approach for product form Kansei design based on text mining and BPNN

Kansei Engineering (KE) is a product design method that aims to develop products to meet users’ emotional preferences. However, traditional KE faces the problem that the acquisition of Kansei factors does not represent the real consumers demands based on manual and reports, and using traditional methods to calculate relationship between Kansei factors and specific design elements, which can lead to the omission of key information. To address these problems, this study adopts text mining and backward propagation neural networks (BPNN) to propose a product form design method from a multi-objective optimization perspective. Firstly, Term Frequency-Inverse Document Frequency (TF-IDF) and WordNet are used to extract key user Kansei requirements from online review texts to obtain more accurate Kansei knowledge. Secondly, the BPNN is used to establish the non-linear relationship between product Kansei factors and specific design elements, and a preference mapping prediction model is constructed. Finally, BPNN is transformed into an iterative prediction value of non-dominated sorting genetic algorithm-II (NSGA-II), and the model is solved through multi-objective evolutionary algorithm (MOEA) to obtain the Pareto optimal solution set that satisfies the user’s multiple emotional needs, and the fuzzy Delphi method is used to obtain the best product form design scheme that meets the user’s multiple emotional images. Using the example of electric bicycle form design could show that this proposed method can effectively complete multi-objective product solutions innovation design.

Evolutionary multitasking for multiobjective optimization based on hybrid differential evolution and multiple search strategy

Multitask optimization uses the knowledge transfer between tasks to deal with multiple related tasks simultaneously, which obtains better optimization performance. Currently, most multitask optimization algorithms have shortings such as slow convergence speed, difficulty in generating high-quality solutions and insufficient population diversity. To solve these problems, an new multiobjective multitask evolutionary algorithm is proposed in this paper. The algorithm introduces hybrid differential evolution strategy and multiple search strategy to generate offspring and high-quality solutions. In order to accelerate the population convergence and improve the lack of population diversity, we mixed two differential mutation operators with different functions to generate offspring. While generating high-quality solutions to improve the population convergence speed, some random solutions can also be generated to maintain the population diversity, which enhances the ability of the algorithm to jump out of the local optimal. Secondly, the multiple search strategy uses the information collected from multiple dimensions to optimize decision variables, and generates high-quality solutions by migrating useful knowledge. Finally, the performance of the algorithm is verified on two classical multitask test set. Compared with other related algorithms, the proposed algorithm has a faster convergence speed and better distribution performance, and shows a good ability to jump out of local optimal.

Feature selection for sentiment analysis using hybrid multiobjective evolutionary algorithm

As the volume of data continues to grow, the significance of text classification is on the rise. This vast amount of data majorly exists in the form of texts. Effective data preparation is essential to extract sentiment data from this vast amount of text, as irrelevant and redundant information can impede valuable insights. Feature selection is an important step in the data preparation phase as it eliminates irrelevant and insignificant features from the huge features set. There exist a large body of work related to feature selection for image processing but limited research is done for text data. While some studies recognize the significance of feature selection in text classification, but there is still need for more efficient sentiment analysis models that optimize feature selection and reduce computational. This manuscript aims to bridge these gaps by introducing a hybrid multi-objective evolutionary algorithm as a feature selection mechanism, combining the power of multiple objectives and evolutionary processes. The approach combines two feature selection techniques within a binary classification model: a filter method, Information Gain (IG), and an evolutionary wrapper method, Binary Multi-Objective Grey Wolf Optimizer (BMOGWO). Experimental evaluations are conducted across six diverse datasets. It achieves a reduction of over 90 percent in feature size while improving accuracy by nearly nine percent. These results showcase the model’s efficiency in terms of computational time and its efficacy in terms of higher classification accuracy which improves sentiment analysis performance. This improvement can be beneficial for various applications, including recommendation systems, reviews analysis, and public opinion observation. However, it’s crucial to acknowledge certain limitations of this study. These encompass the need for broader classifier evaluation, and scalability considerations with larger datasets. These identified limitations serve as directions for future research and the enhancement of the proposed approach.

Multi-objective evolutionary algorithm-enabled multi-stage collaborative scheduling for automotive production

In the automotive industry, considering all the process workshops as a whole in terms of production scheduling becomes much more significant for the enhancement and optimization of overall productivity, efficiency, resource utilization, and coordination among factories. However, the complicated operational interdependencies between workshops make it hard to acquire a global objective. This paper aims to model the collaborative scheduling problem for a multi-stage automotive production process first, involving critical decision variables from four main workshops, stamping, welding, painting, and assembling. Then, the multi-objective evolutionary algorithm based on Pareto optimal subspace learning (MOEA/PSL) associated with an encoding and decoding strategy based on a random key is designed to solve the model for minimizing the total cost and weighted tardiness. Finally, a real-life case study is carried out to illustrate the effectiveness and superiority of the proposed method via experimental comparison using practical data and simulated instances for further analysis.

Boosting scalability for large-scale multiobjective optimization via transfer weights

Large-scale multiobjective optimization problems (LSMOPs), which optimize multiple conflicting objectives with hundreds or even thousands of decision variables, demand increasing computational resources to assure satisfactory performance as the decision variables increase. Multiobjective evolutionary algorithms are naturally scalable, but as the dimension increases, the conflict between analyzing enough solutions and the limited number of function evaluations has hindered further improvements in scalability. In this paper, we first define the scalability of multiobjective evolutionary algorithms and design an indicator to quantitatively measure the scalability. Second, to boost scalability when solving LSMOPs, we propose a scalable multiobjective optimization algorithm by transferring weights between solutions to reduce dependency on ever-increasing computational resources as the problem dimension increases. The proposed framework entails constructing a latent decision space to determine evolutionary weights for chosen representative solutions. These computed weights are then transferred to the remaining solutions, enabling evolutionary optimization to proceed without requiring additional evaluations, even as the dimensionality increases. By utilizing the knowledge gained from the source solutions, each solution is customized with an evolutionary weight scheme that not only preserves computational resources but also enhances optimization performance, thereby boosting scalability. We have conducted experiments on LSMOPs to verify the effectiveness and scalability of the proposed algorithm. The proposed method outperforms selected state-of-the-art algorithms and gains scalability boosts in situations where the dimensions increase while the function evaluation remains constant.

Identification of the Constitutive and Friction Models Parameters via a Multi-Objective Surrogate-Assisted Algorithm for the Modeling of Machining—Application to Arbitrary Lagrangian Eulerian Orthogonal Cutting of Ti6Al4V

Abstract The evolution of high-performance computing facilitates the simulation of manufacturing processes. The prediction accuracy of a numerical model of the cutting process is closely associated with the selection of constitutive and friction models. The reliability and the accuracy of these models highly depend on the value of the parameters involved in the definition of the cutting process. Direct of inverse methods are used to determine these model parameters. However, these identification procedures often neglect the link between the parameters of the material and the friction models. This article introduces a novel approach to inversely identify the best parameters value for both models at the same time and by taking into account multiple cutting conditions in the optimization routine. An artificial intelligence (AI) framework that combines the finite element modeling with an adaptive Bayesian multi-objective evolutionary algorithm (AB-MOEA) is developed, where the objective is to minimize the deviation between the experimental and the numerical results. The arbitrary Lagrangian–Eulerian (ALE) formulation and the Ti6Al4V alloy are selected to demonstrate its applicability. The investigation shows that the developed AI platform can identify the best parameters values with low computational time and resources. The identified parameters values predicted the cutting and feed forces within a deviation of less than 4% from the experiments for all the cutting conditions considered in this work.

A hybrid multi-objective evolutionary algorithm with high solving efficiency for UAV defense programming

Emerging Unmanned Aerial Vehicle (UAV) application patterns, including the Loyal Wingman and Unmanned Swarms, have significantly challenged the administration and defense against illegal or trespassing UAVs. The weapon-target assignment (WTA) problem, a famous combinatorial optimization problem in military operations research, is decisive for the success of UAV defense programming. Related investigations face two main challenges: 1) Constructing a multi-objective optimization model reflecting the complexities of new UAV application patterns, and 2) solving the time-consuming WTA problem with high-dimension decision variables as it is NP-complete. This paper introduces a modified differential evolution based on a new mapping method called Pareto-Optimal-Matching that solves multi-objective binary optimization problems and constructs a 3 objective Sensor-Weapon-Target Assignment model. The proposed algorithm overcomes the inherent limitation of traditional differential evolution, which only performs well in continuous problems. Promising results were obtained based on different simulation metrics under various problems (proposed model, MOKP, and ZDT5), validating the superior performance of the proposed algorithm.

A Multi-Task Decomposition-Based Evolutionary Algorithm for Tackling High-Dimensional Bi-Objective Feature Selection

Evolutionary algorithms have been widely applied for solving multi-objective optimization problems, while the feature selection in classification can also be treated as a discrete bi-objective optimization problem if attempting to minimize both the classification error and the ratio of selected features. However, traditional multi-objective evolutionary algorithms (MOEAs) may have drawbacks for tackling large-scale feature selection, due to the curse of dimensionality in the decision space. Therefore, in this paper, we concentrated on designing an multi-task decomposition-based evolutionary algorithm (abbreviated as MTDEA), especially for handling high-dimensional bi-objective feature selection in classification. To be more specific, multiple subpopulations related to different evolutionary tasks are separately initialized and then adaptively merged into a single integrated population during the evolution. Moreover, the ideal points for these multi-task subpopulations are dynamically adjusted every generation, in order to achieve different search preferences and evolutionary directions. In the experiments, the proposed MTDEA was compared with seven state-of-the-art MOEAs on 20 high-dimensional classification datasets in terms of three performance indicators, along with using comprehensive Wilcoxon and Friedman tests. It was found that the MTDEA performed the best on most datasets, with a significantly better search ability and promising efficiency.

Cooperative constrained multi-objective dual-population evolutionary algorithm for optimal dispatching of wind-power integrated power system

Due to increasing uncertainty and multiple operational requirements of current power system, it brings great challenge to power dispatch of wind-power integrated power system. To address above problems, a cooperative constrained multi-objective dual-population evolutionary algorithm (CCMDEA) is proposed with uncertainty budget of wind power, which divides the intermittent power into different intervals with flexible uncertain parameters. On the basis of non-dominated sorting genetic algorithm III (NSGA-III), constraint violation (CV) is also involved to guide the population evolution combined with two defined populations: convergence population (CP) and diversity population (DP) to coordinate convergence and diversity performance of pareto-fronts, and a coordinated selection mechanism is designed to select suitable parents according to the state of evolutionary population, which can coordinate the convergence population and diversity population well. According to simulation results on two test systems, the efficiency and priority of the proposed algorithm is well verified, which also reveals that the proposed algorithm can be a viable way for solving power dispatching of wind-power integrated power system.

Dual population multi-objective evolutionary algorithm for dynamic co-transformations

Dual population multi-objective evolutionary algorithm for dynamic co-transformations

Multi-objective evolutionary algorithm with evolutionary-status-driven environmental selection

The hardly dominated boundary (HDB) is a common feature of multi-objective optimization problems (MOPs). Previous studies have proposed several multi-objective evolutionary algorithms (MOEAs) to deal with the problem characterized by HDBs. Nevertheless, these methods lack consideration of the evolutionary status, potentially resulting in low efficiency when addressing problems that entail a confluence of HDB and other intricate characteristics. This paper proposes an MOEA with evolutionary-status-driven environmental selection (MOEA-ESD) to address such an issue. Specifically, in each generation, the current and historical information are used to evaluate the evolutionary status. Based on the estimated evolutionary status, the proposed MOEA-ESD algorithm adaptively uses three types of environmental selection: the traditional environmental selection (TES) based on Pareto-dominance and crowding distance, a convergence-first environmental selection (CFES) based on Gaussian mixture model clustering, and a diversity-first environmental selection (DFES) based on outlier detection and extreme solutions. In this way, inferior solutions situated on the HDB can be effectively eliminated, thereby facilitating the convergence of the population while concurrently preserving a commendable degree of diversity. Moreover, a set of new benchmark problems with different objective magnitudes and complicated Pareto sets is developed to enrich the features of HDB-MOPs and to verify the algorithm's performance. Our experimental results on 22 HDB-MOPs show the promising performance of the proposed algorithm. The source code of MOEA-ESD is available at https://github.com/CIAM-Group/EvolutionaryAlgorithm_Codes/tree/main/MOEA-ESD.

Multi-objective optimization of sustainable cement-zeolite improved sand based on life cycle assessment and artificial intelligence

Background Cement-zeolite improved sand can be used in diverse civil engineering applications. However, earlier research has not duly optimized its production process to attain best mechanical strength, lowest cost, and least environmental impact. This study proposes a multi-objective optimization approach using back-propagation neural network (BPNN) to predict the mechanical strength, along with an adaptive geometry estimation-based multi-objective evolutionary algorithm (AGE-MOEA) to identify the best parameters for cement-zeolite-improved sand, filling a long-lasting research gap. Methods A collection of unconfined compression tests was used to evaluate cemented sand specimens treated with stabilizers including portland cement (at dosages of 2, 4, 6, 8, and 10%) and six dosages of natural zeolite as partial replacement for cement (0, 10, 30, 50, 70, and 90%) at different curing times of 7, 28, and 90 days. The study further conducts a detailed analysis of life cycle assessment (LCA) to show how partial zeolite replacement for cement impacts the environment. Through a tuning process, the BPNN model found the optimal architecture and accurately predicted the unconfined compressive strength of cement-zeolite improved sand systems. This allowed the AGE-MOEA to optimize zeolite and cement dosages, density, curing time, and environmental impact. Results The results of this study reveal that the optimal range of zeolite was between 30-45%, which not only increased cemented sand strength, but also reduced the cost and environmental impact. It is also shown that increasing the zeolite replacement to 25-30% can increase the ultimate strength of cemented sand, yet exceeding this limit can cause the strength to decrease. Conclusions Zeolite has the potential to serve as an alternative for cement in applications that involve cemented sand, while still achieving mechanical strength performance, which is comparable or even superior. From an LCA standpoint, using zeolite as partial cement replacement in soil improvement projects is a promising alternative.

Risk control of epidemic in urban cold-chain transportation

The COVID-19 pandemic has severely disrupted the daily running of urban logistics system, thereby increasing the resilience and health requirements in the design of sustainable logistics system. Most scientific contributions have enriched the knowledge of how to deal with viral infection threats from the external environment but ignored the virus transmission from fomites within the logistics system. This study designs an urban cold-chain logistic network with control strategies to reduce the epidemic risk caused by cold-chain fomites while concurrently balancing the total cost and customer satisfaction, which is modeled as a multi-objective vehicle routing problem (MOVRP). The control strategies are modeled by dividing customer points into small groups, and the virus transmission in populations is considered when evaluating epidemic risk. A multi-objective evolutionary algorithm with special initialization (MOEA-SI) is proposed to solve this problem. Simulation results show that all the strategies can efficiently reduce the epidemic risk with reasonable expense of total cost and customer satisfaction. The introduction of special initialization strategy enables the MOEA-SI to have good performance in balancing the three objectives. The above results can enrich the knowledge of how to deal with unexpected epidemics and help the scientific decision-making in sustainable urban management.

A matheuristic-based multi-objective evolutionary algorithm for flexible assembly jobs shop scheduling problem in cellular manufacture

The flexible assembly job shop scheduling problem (FAJSP) and its extensions have received increased attention. However, multi-objective FAJSP in cellular manufacture (FAJSP-CM), a real-world extension of FAJSP with important applications in the mass customization of complex products, is barely considered in most previous studies. Hence, this study first addresses the multi-objective FAJSP-CM problem via a mixed-integer linear programming model (MILP). Then, a matheuristic-based multi-objective evolutionary algorithm (MMOEA), with a matheuristic decoding method, knowledge-guided initialization and learning-based local intensification method, is presented to efficiently tackle the FAJSP-CM problem in polynomial-time. The matheuristic decoding method, which hybrids a decomposition linear programming model (D-LP) that can quickly and optimally allocate the execution times of all operations, to search for the optimal solution in a specific space. The knowledge-guided initialization is extended to get a high-quality and great-diversity initial population for accelerated convergence. Meanwhile, the learning-based local intensification method is developed to adaptively coordinate five problem-specific local search operators to enhance the exploitation ability of the MMOEA algorithm. Finally, comprehensive experiments based on 720 instances are conducted to evaluate the proposed MILP model and MMOEA algorithm. The results of the first experiments illustrate the MILP model can obtain the Pareto solutions of small-scale instances. The remaining experiments demonstrate that the proposed MMOEA algorithm outperforms its competitors by 0.21 and 0.72 in hypervolume ratio and coverage indictors, along with approximately a 25 % improvement in success rate.

A recursive framework for improving the performance of multi-objective differential evolution algorithms for gene selection

Gene selection is a pivotal process in machine-learning-driven medical diagnostics, where the goal is to identify a subset of genes from microarray expression profiles that can enhance the predictive accuracy of classifiers for disease diagnosis. The two key objectives of gene selection are to reduce the dimensionality of the data and to improve the accuracy of disease diagnosis, which is typically a multi-objective optimization problem. In recent years, multi-objective evolutionary algorithms (MOEAs) have gained wide attention in feature selection research, and several related algorithms have been produced. However, most algorithms tend to get stuck in local optimality when searching for solutions from a high-dimensional space. To solve the gene selection problem effectively, this study introduces a recursive multi-objective differential evolution algorithm with elite recursive strategy (RMODE-E) and a recursive multi-objective differential evolution algorithm with Pareto front recursive strategy (RMODE-P). RMODE-E amalgamates the features selected by the top E elite individuals, RMODE-P consolidates the features selected by the Pareto front set, and the combined features then serve as the foundation for subsequent recursive rounds of searching. The proposed feature subspace combination strategy not only reduces the recursive search space but also improves the capacity to find globally optimal feature subsets. Extensive experiments were conducted to compare our proposed algorithms with eight state-of-the-art evolutionary algorithms to validate their effectiveness. Experimental results demonstrate that RMODE-P has better global search capability as it achieves better best classification accuracy, mean classification accuracy, and minimal gene subset size.

Configuration Optimization of Mobile Photovoltaic-Diesel-Storage Microgrid System Based on CPS-MOEA

This paper presents a two-step approach for optimizing the configuration of a mobile photovoltaic-diesel-storage microgrid system. Initially, we developed a planning configuration model to ensure a balance between the mobility of components and a sustainable power supply. Then, we introduced a method that merges optimization and decision-making. The first phase identifies Pareto optimal solutions (POSs) with a favorable distribution by using a multi-objective evolutionary algorithm with classification-based preselection (CPS-MOEA). In the second phase, we utilize the fuzzy C-means algorithm (FCM) and the grey relational projection (GRP) method for comprehensive decision-making. This aims to select the most suitable and compromise solution from the POSs, closely aligning with the decision-maker’s preferences. Beyond addressing the optimal planning and configuration issue, the experimental results show that the method surpasses other widely used multi-objective optimization algorithms, including the Preference Inspired Co-evolution Algorithm (PICEA-g), the Multi-Objective Particle Swarm Optimization Algorithm (MOPSO), and the third stage of Generalized Differential Evolution (GDE3).